The present invention relates to an imaging apparatus detecting a face portion included in image data, and more particular to an imaging apparatus changing the detection method of the face portion according to an operation state, a processing method in each imaging apparatus and a program making a computer execute the method.
In recent years, a digital still camera capable of imaging not only an original still image but also a moving image has rapidly spread. These digital still cameras have become highly efficient, and especially detection/recognition system techniques have been watched. The representative technique of the detection/recognition system techniques is a face detection technique.
Now, when a subject is imaged with a digital still camera, imaging is frequently performed with the digital still camera inclined in various directions. Imaging is frequently performed with a digital camera long from side to side rotated by “90 degrees”, for example. As to such rotations of a digital still camera, a “0 degree” rotation, a “+90 degree” rotation, a “−90 degree” rotation and a “180 degree” rotation are supposed when a clockwise rotation is supposed to be a positive direction. Conventionally, as to the rotation of the digital still camera, a technique of inputting the information related to the direction of a rotation, for example a “+90 degree” rotation, to detect a face portion using the image data obtained by an imaging based on the input rotation direction has been proposed (see, for example, Patent Document 1). Moreover, a technique of counting the number of face portions when there is a plurality of face portions detected from image data, to use the number for extra copies of a photograph has been also proposed (see, for example, Patent Document 2).
[Patent Document 1] Japanese Patent Application Publication No. 2004-62565 (at FIG. 1)
[Patent Document 2] Japanese Patent Application Publication No. Hei 9-281605 (at FIG. 1)
However, the above technique of inputting the information of the direction of a rotation to detect a face portion in the image data imaged based on the input rotation direction requires the previous input of the rotation direction, and consequently the operation is troublesome. For detecting the face portion from the image data without performing such an operation, it is conceivable to search for the face portion of the image data from all of the rotation directions in consideration of all cases of imaging after “0 degree” rotation, “+90 degree” rotation, “−90 degree” rotation and “180 degree” rotation. But that method requires plenty of time, and is not realistic. On the other hand, at the time of searching for image data in face detection, if, for example, only the “0 degree” rotation and the “+90 degree” rotation are performed without considering all the cases of the imaging after the “0 degree” rotation, the “+90 degree” rotation, the “−90 degree” rotation and the “180 degree” rotation, then it would be possible to perform high speed face detection. But, in that case, the case where no face portions can be detected may increase, and it is apprehended that a detection rate decreases.
Moreover, if the information obtained at the time of detecting a face portion from image data can be recorded as the information accompanying the image of the image data in association with the image data, it becomes possible to retrieve already recorded image data from various viewpoints.